.png)
Frameworks, core principles and top case studies for SaaS pricing, learnt and refined over 28+ years of SaaS-monetization experience.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
How Could ServiceNow's AI Agent Fabric Reshape Its Pricing Power
ServiceNow has spent two decades becoming the workflow layer through which large enterprises manage incidents, employee requests, security events, customer cases, assets, approvals, and change. AI Agent Fabric raises the ambition. Instead of orchestrating work performed mainly by people, ServiceNow now wants to coordinate work performed by thousands of its own agents, customer-built agents, and third-party agents from companies such as Microsoft, Google Cloud, Adobe, Cisco, Box, and IBM. Launched on 6 May 2025, the fabric supports agent-to-agent and agent-to-system communication through common protocols such as MCP and A2A.
The commercial stakes are larger than another premium software module. ServiceNow reported on 22 July 2026 that its AI products had passed $1 billion in annual contract value, while agentic deployments had increased ninefold in nine months. Management expects AI to provide 30% of company ACV by 2030. The central pricing question is therefore no longer how much to charge for an AI assistant. It is how ServiceNow should capture value when autonomous systems use its platform to execute work across the enterprise.
Monetizely’s position is that Agent Fabric will strengthen ServiceNow’s pricing power only if autonomous work stops being treated mainly as a tier entitlement. ServiceNow should retain an annual platform commitment as its primary meter, include a defined allowance of completed agent actions, and charge transparent overages for additional cross-system actions - not for seats, tokens, or vaguely defined AI capacity.
Agent Fabric turns workflow ownership into enterprise-wide coordination
ServiceNow’s traditional advantage came from owning the workflow. An incident began in ServiceNow, moved through its rules and approvals, and ended in a ServiceNow record. Agent Fabric expands that role. A Microsoft agent might identify an employee’s need, a ServiceNow agent could determine the required workflow, a Box agent could retrieve a document, and an SAP or Workday process could execute the final transaction. ServiceNow would coordinate the chain even when much of the intelligence came from elsewhere.
That distinction matters because coordination can become more valuable than any one agent. Rao Surapaneni, Google Cloud’s vice-president for business applications, said in May 2025 that agents “must operate seamlessly across diverse applications, data, and clouds”. Amit Zavery, ServiceNow’s president and chief product officer, compared coordinating agents with “leading human employees”.
Monetizely’s 5-Step Pricing Framework begins with Goals and Segmentation, which clarifies what the company wants pricing to achieve and which buyers it serves. Packaging then groups capabilities into offers that fit those segments. Pricing Metric determines what the customer pays for, such as a user, action, workflow, or outcome. Rate Setting establishes the actual prices, discounts, and volume curves. Operationalisation makes the design work through entitlements, metering, rating, billing, revenue recognition, and customer reporting. The sequence matters because no clever rate can repair a package that serves the wrong buyer or a metric that fails to track value. The same logic is developed for autonomous products in Monetizing Agentic AI, where products must connect revenue not only to access but also to the work agents perform.
ServiceNow’s pricing evolution shows that its product has moved faster than its commercial model.
| Period | Product and packaging structure | Commercial signal |
|---|---|---|
| By January 2025 | Now Assist for ITSM required ITSM Pro Plus or Enterprise Plus, placing generative AI behind premium product tiers. | AI was packaged as a higher-tier feature intended to increase subscription ACV. |
| May 2025 | AI Control Tower launched as the governance layer, while AI Agent Fabric connected ServiceNow, customer-built, and third-party agents. | ServiceNow began selling control over an agent ecosystem, not simply productivity within one workflow. |
| December 2025 | ServiceNow stated that certain AI and data products included consumption components when customers exceeded fixed service credits. | Usage pricing existed, but mainly through negotiated credits embedded within subscriptions. |
| August 2026 | ITSM Foundation, Advanced, and Prime each include increasing amounts of AI; Prime contains independent AI specialists and broader autonomous agents. Public list prices remain unavailable. | Agentic maturity now defines the tiers, while the economic unit behind autonomous work remains hidden. |
Sources: ServiceNow product documentation, archived pricing pages, current pricing pages, launch materials, and SEC filings.
The table shows a sound product transition but an unfinished pricing transition. ServiceNow has moved from charging more for AI features to coordinating digital work across vendors, yet customers still lack a clear public unit that explains how the value and cost of that work scale.
The Agentic Monetization Spectrum, or AMS, helps identify the appropriate pricing metric through three dimensions. Zero-human ability measures how independently the agent can work. Operational domain measures whether it handles one task, a complete function, or processes spanning several departments. Output/cost ratio compares the value of completed work with the compute and operating cost required to produce it. Greater autonomy, broader reach, and a steeper output-to-cost ratio move the sound pricing anchor away from human seats and towards work performed.
Agent Fabric does not produce one uniform outcome. It enables other agents to exchange context, trigger actions, apply policies, and complete workflows. Its AMS position therefore points towards a platform-plus-action model rather than either pure seat pricing or pure outcome pricing.
| AMS dimension | ServiceNow Agent Fabric score | Evidence and pricing implication |
|---|---|---|
| Zero-human ability | 4 of 5 | Agents can route, decide, act, and execute tasks, although regulated work still requires policies and human approval. Revenue should rise with autonomous work, not human logins. |
| Operational domain | 5 of 5 | The fabric spans IT, HR, CRM, security, data, employee service, and third-party systems. A narrow per-resolution metric would not cover that range. |
| Output/cost ratio | 4 of 5 | One coordinated action can remove manual routing, system switching, and administrative labour at low marginal compute cost. Attribution remains less direct than in a customer-support resolution. |
| AMS implication | Platform commitment plus completed actions | The platform commitment pays for governance, availability, data, and control. Completed-action charges capture expanding autonomous work. |
The AMS score rules out continuing to treat the seat as the main growth meter. ServiceNow can still charge for fulfiller access where people operate the platform, but the company’s next source of expansion is the number and value of actions coordinated across the customer’s estate.
Customer evidence already points in that direction. Michael Gray, chief technology officer at Thrive, said in May 2025 that ServiceNow automation had routed more than 315,000 tasks and saved 21,000 hours over six months. Suresh Vittal, UKG’s chief product officer, described a target in which employees self-serve 80% of common HR requests. Neither value story is naturally explained by the number of human users who log in.
Pure outcome pricing would go too far, however. A Fin resolution can be defined and observed within one conversation. Agent Fabric may participate in a chain involving six systems, three agents, an approval, and a human decision. Crediting ServiceNow with the final revenue outcome or cost reduction would create disputes over attribution.
A completed action is more defensible. Examples include updating a customer record, opening a security case, applying an approved configuration, issuing an access right, scheduling a technician, or completing an employee request. Every event can be logged, audited, reversed, and connected to the agent and policy that authorised it.
Strong packaging is being weakened by an opaque metric
ServiceNow’s current ITSM packaging recognises three distinct levels of AI adoption. Foundation supplies task-based support and basic AI. Advanced adds agentic workflows and voice agents that operate beside employees. Prime adds AI specialists that independently manage IT workflows and make decisions.
That is materially better than adding one undifferentiated “AI” module to every contract. Buyers can identify whether they need assistance, workflow automation, or autonomous operation.
The difficulty lies in what happens after the package is selected. ServiceNow’s public page directs buyers to request a custom quote. Its 2025 Form 10-K says AI and data products may include service credits and consumption charges, but it does not disclose the unit, conversion rate, included quantity, or overage curve.
Our scorecard separates the quality of the product architecture from the quality of the current commercial design.
| Framework step | Grade | One-line diagnosis |
|---|---|---|
| Packaging | B | Foundation, Advanced, and Prime map well to increasing agent maturity, but concentrating independent agents in Prime may slow wider adoption and testing. |
| Pricing metric | C+ | Fixed commitments protect predictability, yet opaque service credits prevent buyers from connecting incremental spend with completed work. |
| Operationalisation | A- | ServiceNow has strong governance, permissions, workflow records, auditability, and existing credit infrastructure, but customer-facing metering remains difficult to inspect. |
The scorecard means ServiceNow’s pricing weakness is not a lack of technical plumbing. The company already records transactions, governs agents, manages entitlements, and supports consumption components. The problem is that the buyer cannot see a stable relationship between the bill and the work performed.
ServiceNow gets three important things right.
First, it protects the annual platform commitment. Subscription contracts generally run for 12 to 36 months, usually contain fixed consideration, and are normally non-cancellable. That structure produced $12.9 billion in 2025 subscription revenue and $29 billion of remaining performance obligations by June 2026. Abandoning committed revenue for fully variable billing would weaken one of the strongest financial models in enterprise software.
Second, it packages governance with execution. AI Control Tower can discover, monitor, secure, and measure ServiceNow and third-party agents. Agent Fabric then allows those agents to communicate and act. Buyers are not merely purchasing inference. They are buying accountability for what autonomous software does.
Third, ServiceNow can sustain enterprise-scale expansion. The company had 658 customers with more than $5 million in ACV at 30 June 2026, up about 23% year over year, and completed 123 transactions above $1 million in net new ACV during the quarter. Agent Fabric can deepen those relationships because governance becomes more valuable as each customer adds agents from other vendors.
What ServiceNow gets wrong is equally clear. Prime bundles several distinct sources of value - autonomous IT work, Now Assist, Moveworks, DevOps capabilities, DEX agents, and platform functions - into one negotiated offer. Procurement teams cannot tell whether a price increase pays for better software access, more agent activity, additional models, or larger workflow volume.
Opacity can support price discrimination while a product is new. Over time, it makes adoption harder. A CIO deciding whether to connect 20 third-party agents needs a budget rule that can survive scrutiny from finance. “Contact sales” is not a scaling rule.
Public consumption meters are resetting buyer expectations
ServiceNow does not set pricing in isolation. Buyers compare its proposal with the public structures offered by other large platforms, even when the products differ.
By May 2025, Salesforce had moved Agentforce from $2 per conversation to an action model. It sold 100,000 Flex Credits for $500, with a standard action consuming 20 credits, or $0.10. Salesforce also introduced a digital wallet and allowed customers to convert spending between user licences and digital labour.
Microsoft introduced pay-as-you-go Copilot Studio pricing in December 2024 at $0.01 per message. An autonomous action consumed 25 messages, while more complex use could consume additional units. The unit may not be elegant, but it is inspectable.
Customer-service vendors have gone further towards outcomes. Intercom charges $0.99 when Fin delivers a defined outcome, HubSpot moved Breeze Customer Agent to $0.50 per resolved conversation in April 2026, and Zendesk now meters AI agents through automated resolutions rather than its former monthly-active-user structure.
The comparison supports a broader pricing conclusion.
| Vendor and date | Primary AI meter | Buyer can forecast marginal use? | Lesson for ServiceNow |
|---|---|---|---|
| ServiceNow, August 2026 | Custom subscription with included credits and negotiated consumption | Partly | Strong commitment model, weak public connection between unit and work |
| Salesforce, May 2025 | $0.10 per standard Agentforce action | Yes | Actions can span functions better than conversations or seats |
| Microsoft, December 2024 | $0.01 per message, with weighted consumption for autonomous work | Yes | Complexity can be reflected through published unit conversion |
| Intercom, June 2026 | $0.99 per defined Fin outcome | Yes | Clear success definitions can remove payment risk |
| HubSpot, April 2026 | $0.50 per resolved conversation or $1 per recommended lead | Yes | Different agents can carry different outcome prices |
| Zendesk, June 2026 | Automated resolutions layered onto support plans | Partly | Access and autonomous output can coexist in one contract |
ServiceNow should not copy the customer-support outcome meter. Its domain is too broad. The important competitive lesson is transparency: Salesforce, Microsoft, Intercom, HubSpot, and Zendesk have all given customers a visible way to connect another unit of agent work with another unit of spend.
ServiceNow’s own financial disclosures confirm that management understands the transition. Its 2025 10-K says the company may add more consumption-based components, while warning that customers might reject pricing changes. That risk is real. Poorly defined consumption pricing can feel like a cloud bill that grows faster than the value received.
The answer is not to conceal usage. Better design makes it controllable.
A completed-action layer can raise ACV while preserving budget control
ServiceNow’s next pricing reset should have one coherent architecture.
The annual governed-platform commitment remains the primary meter. It pays for AI Control Tower, Agent Fabric availability, security, audit records, policy enforcement, data connections, model choice, support, and a defined amount of agent activity.
Above that base, customers pay for completed agent actions. A billable action occurs when an authorised agent writes, updates, approves, launches, closes, routes, or changes something in an enterprise system. Failed attempts, reads, internal reasoning steps, and reversed transactions should not be billed.
A public conversion table can weight work by complexity. A simple record update might consume one unit, a multi-system action five units, and an autonomous workflow completed without human intervention 20 units. Customers would see usage and forecast demand through an account-level wallet, with departmental budgets, alerts, spend caps, and a full action log.
The model below shows how the structure could improve ServiceNow’s revenue capture without forcing the customer to accept an uncapped variable bill.
| Modelled enterprise account | Year one | Year two | Year three | Three-year total |
|---|---|---|---|---|
| Current fixed bundle, with 5% annual uplift | $2.00m | $2.10m | $2.21m | $6.31m |
| Governed-platform commitment | $1.85m | $1.94m | $2.04m | $5.83m |
| Completed-action overage | $0.20m | $0.40m | $0.70m | $1.30m |
| Recommended total | $2.05m | $2.34m | $2.74m | $7.13m |
| Increase over current structure | 2.5% | 11.5% | 24.2% | 13.1% |
The synthesis is straightforward: ServiceNow accepts slightly less guaranteed platform revenue at the start, but earns more as autonomous activity grows. The customer gains an auditable cost curve rather than facing a large tier jump or an opaque credit true-up.
Such a reset also protects gross margin. ServiceNow’s non-GAAP subscription gross margin was 80.5% in Q2 2026, and management said faster AI adoption and greater use of hyperscaler partners affected its gross-margin outlook. A completed-action layer allows rates and included allowances to reflect inference, cloud, orchestration, and support costs without exposing customers to tokens.
Model efficiency should reinforce that position. ServiceNow and NVIDIA announced a purpose-built 15-billion-parameter model in May 2025 intended to lower latency and inference expense for enterprise agents. ServiceNow can therefore price actions according to customer value while using model routing and specialised models to reduce the cost of delivery.
Most importantly, the reset creates a reason for ServiceNow to welcome third-party agents rather than defend only its own. Every Microsoft, Google, Salesforce, Box, or customer-built agent that performs governed work through ServiceNow could increase action volume. Agent Fabric would turn interoperability into revenue.
That is the source of the new pricing power. ServiceNow need not own every agent. It needs to own the trusted point at which agents receive authority, access enterprise context, take action, and leave an auditable record.
Pricing power will come from becoming the control layer buyers can budget
ServiceNow entered 2026 with considerable momentum. Q2 subscription revenue reached $3.877 billion, AI ACV crossed $1 billion, and management reported ninefold growth in agentic deployments over nine months. Those figures show that enterprises will already pay a premium for the platform’s AI capabilities.
The next phase requires pricing to catch up with the product. Tiered subscriptions remain useful for segmenting buyers, but autonomous work cannot remain an invisible feature inside Prime. Nor should ServiceNow copy a narrow resolution fee suited to customer service. Its product coordinates actions across functions, vendors, and systems, so its billable unit should reflect that role.
Monetizely’s committed view is that ServiceNow should make the platform commitment primary and completed agent actions secondary. Governance earns the commitment. Work performed earns the consumption revenue. Together, they allow ServiceNow to expand ACV as digital labour grows without asking customers to buy more human seats.
Executives evaluating ServiceNow or designing a comparable agent platform should take four actions:
Model the company around digital-work growth, not licence growth. Board plans should track completed autonomous actions and governed third-party agents alongside ACV, renewals, and user counts.
Treat interoperability as a revenue strategy. Make it economically attractive for outside agents to execute through the platform, even when they compete with native agents.
Separate AI adoption from premium-tier migration. Allow customers to begin with a meaningful action allowance before requiring a broad suite upgrade, then let demonstrated activity drive expansion.
Make trust a paid platform service. Position policy enforcement, auditability, identity, security, and action reversal as the reason for the annual commitment rather than presenting them as technical overhead.
ServiceNow’s opportunity is larger than selling better workflow software. Agent Fabric can make the company the operating layer for a mixed workforce of people and agents. The firms that control that layer will not merely charge for software access. They will participate economically in the work the enterprise completes.
Assumptions
The three-year account model assumes a $2 million current annual contract, 5% yearly price growth, a proposed $1.85 million platform commitment, five million included actions per year, a $0.04 overage rate, and completed-action volumes of 10 million, 15 million, and 22.5 million. The model excludes implementation, professional services, taxes, currency changes, negotiated discounts, and infrastructure pass-through charges.
Footnotes
https://www.amazon.com/Monetizing-Agentic-AI-Handbook-Transformation/dp/B0H7Z13VKJ/
https://newsroom.servicenow.com/press-releases/details/2025/ServiceNow-Launches-AI-Control-Tower-a-Centralized-Command-Center-to-Govern-Manage-Secure-and-Realize-Value-From-Any-AI-Agent-Model-and-Workflow/
https://www.sec.gov/Archives/edgar/data/1373715/000137371526000007/now-20251231.htm
https://investor.servicenow.com/news/news-details/2026/ServiceNow-Reports-Fourth-Quarter-and-Full-Year-2025-Financial-Results-Board-of-Directors-Authorizes-Additional-5B-for-Share-Repurchase-Program/default.aspx
https://investor.servicenow.com/news/news-details/2026/ServiceNow-Reports-Second-Quarter-2026-Financial-Results/default.aspx
https://www.servicenow.com/products/itsm/pricing.html
https://www.servicenow.com/docs/r/xanadu/release-notes/rn-summary-addtl-reqs.html
https://web.archive.org/web/2025/https://www.servicenow.com/products/itsm/pricing.html
https://www.servicenow.com/jp/company/media/press-room/nvidia-enterprise-ai-agents.html
https://www.salesforce.com/in/news/press-releases/2025/05/15/agentforce-flexible-pricing-news/
https://www.microsoft.com/en-us/microsoft-copilot/blog/copilot-studio/whats-new-in-copilot-studio-january-2025/
https://www.intercom.com/help/en/articles/8205718-fin-ai-agent-outcomes
https://www.hubspot.com/company-news/hubspots-customer-agent-and-prospecting-agent-now-you-pay-when-the-task-is-complete
https://support.zendesk.com/hc/en-us/articles/6931689272090-Moving-to-automated-resolutions-from-existing-pricing-plans